Hey,

I've been using TMG array for last 2-3 years and noticed few weird things. I see the more TMGs I add to the array, the more weird it becomes.. TMG is used mainly for Exchange publishing, therefore there any lots of connections and since TMG is handling the authentication, once in a while, during peak times, it fails to authenticate all requests. Probably due to LDAP limitations.

I had data center move recently and one TMG server failed. I had only four TMG server in the array and I noticed that it was performing much better, than before. Before it was Five TMG server in the arrays. Only few hundred auth errors happened during peak times.

I recovered 5th server and added it back to the array, and I see, old results. Up to the 1000 errors happened during first day.

It seems to me that TMG array of 4 members was working much better than TM away of 5 members..

Any ideas why?

Hi

Thank you for your post here.

Do you use unicast or multicast mode? If is unicast mode, it may results in flood and performance may be reduced.

And you can carefully check the authentication failed errors, we can try to analyze the problem depend on these errors.

You also can go to run performance monitor to observe TMG.

Every TMG has updated to the latest version? And All TMG has the same version?

If the detailed information is not enough, it is hard to locate a performance issue.

Best Regards

Quan Gu

Hello,

Thank you for response.

So far things are not that bad, but last week and this week are light. Many customers are out and do not care about the emails.

So far I've got only few errors during some peak times. Majority of errors are: 10013, 10048:

-=-=-

It used to be thousands of those errors a day and many angry customers, but it is better now. I had case open with Microsoft and we did lots of tweaking at Windows Winsock level.

I wish to eliminate this error, but I can live with few hundred per day..

All TMG servers are power full servers, 4 core, enough memory, latest OS, TMG patches.